Background. The prevalence of diabetes in Alabama has increased to 10.5 %, one of the highest among all states. We propose a 4-year randomized trial of an Internet-based intervention for improving guideline adherence by rural Alabama physicians caring for adult patients with type 2 diabetes. Rural Alabama has many disadvantaged and minority patients with limited health care access. Patients with diabetes in rural Alabama receive poorer quality of care and have worse outcomes compared to urban patients. Objectives. (1) Assess barriers to implementation of diabetes guidelines and identify solutions through focus groups and case-based vignette surveys; (2) Develop and implement an interactive Internet intervention including individualized physician performance feedback; (3) Evaluate the intervention in a randomized controlled trial; and (4) Examine the sustainability of improved guideline adherence once feedback ceases. Methods. In partnership with the University of Alabama rural medicine program, we will randomize 200 rural physician offices to an intervention or comparison arm. Our 18-month intervention, customized to the individual physician in real-time, consists of Internet learning modules with case-based education, performance feedback, and benchmarks. The comparison group will receive a text-based, non-interactive Internet posting of publicly available resources. Nurse practitioners and physician assistants from the offices of study physicians may also participate in the Internet modules. Outcomes will be based on previously developed and validated quality measures for diabetes. The intervention will cover screening, diagnosis, treatment, and prevention. Performance feedback will include 15 charts per intervention physician. The main analysis, conducted at the physician level, will compare differential improvement in guideline adherence between the study arms. Ancillary analyses will examine the effects of physician characteristics, other providers in the office, and patient characteristics (e.g., comorbidities, ethnicity, gender, age, and socioeconomic status). Multivariable techniques will adjust for repeated measures, clustering of patients within physicians, and multiple providers within a single office. Significance. This study offers a technologically advanced, theory-grounded intervention for improving care of a high-risk, underserved population. With expertise in translating research into practice, rural medicine, behavioral medicine, health informatics, and clinical diabetes, our multidisciplinary team has a proven record of collaboration. This project will produce an evidence-based and replicable intervention that can be sustained in the "real world," and easily modified for other diseases. This project is substantially improved after making important changes recommended in the second review.